Online Fault Detection and Isolation Method Based on Belief Rule Base for Industrial Gas Turbines

Document Type : Research Paper

Authors

1 Department of Industrial Engineering Iran University of Science and Technology Tehran/Iran

2 Department of Industrial Engineering Iran University of Science and Technology, Tehran, Iran.

3 Department of Industrial Engineering, Iran University of Science and Technology, Tehran, Iran

10.22070/jqepo.2020.4114.1097

Abstract

Real time and accurate fault detection has attracted an increasing attention with a growing demand for higher operational efficiency and safety of industrial gas turbines as complex engineering systems. Current methods based on condition monitoring data have drawbacks in using both expert knowledge and quantitative information for detecting faults. On account of this reason, this paper proposes an online fault detection and isolation method based on belief rule base (BRB), which can deal with modeling behavior of complex systems when semi-quantitative information is available. Although it is difficult to obtain accurate and complete quantitative information, some expert knowledge can be collected and represented by a BRB, which is essentially an expert system. As such, a new BRB based diagnosis model is proposed to detect and isolate faults of the system in real-time when aiming at isolating various damages and determining the severity of each. Moreover, a recursive algorithm is developed for online updating the parameters of the fault diagnosis model. Equipped with the recursive algorithm, the proposed diagnosis model can determine the severity of fault in real-time when two types of faults are dependent and competitive. To prove its potential application, experimental results demonstrate that the proposed model can track the fault severity very well, and the faults can be diagnosed accurately in real time. Thus, R2 values of 0.99782 and 0. 99782 were obtained for the fault estimation of fouling and erosion, respectively, indicating the accurate performance of the proposed model.

Keywords

Journal of Quality Engineering and Production Optimization
Volume 4, Issue 2
May 2019
Pages 49-66

Files

Share

How to cite

Statistics